# Development and validation of the AUDEXCEL algorithm as a diagnostic tool for occupational noise-related hearing disorder

## TL;DR

The paper introduces AUDEXCEL, an Excel-based algorithm validated for diagnosing occupational noise-related hearing disorder in Malaysian workers.

## Contribution

AUDEXCEL is a novel Excel-based diagnostic algorithm for occupational noise-related hearing disorder with high validity and reliability.

## Key findings

- AUDEXCEL achieved 100% sensitivity, specificity, PPV, and NPV for normal hearing and hearing loss categories.
- For noise-induced hearing loss, AUDEXCEL showed 90.3% sensitivity and 98.0% specificity with almost perfect agreement.
- The algorithm demonstrated high reliability (Cohen’s kappa coefficients of 1.000 to 0.922) across all diagnostic categories.

## Abstract

Occupational noise-related hearing disorder (ONRHD) is an occupational disease that poses a significant challenge for workers both globally and locally. To address this, the study aimed to develop and validate AUDEXCEL, an Excel-based algorithm designed to diagnose ONRHD among workers in Malaysia.

This cross-sectional study involved audiograms from 320 workers. These audiograms were first analyzed by experienced occupational health doctors (OHDs) to establish a gold standard diagnosis. The same audiograms were then assessed using the AUDEXCEL algorithm, which was validated through 5-fold cross-validation. Validity (sensitivity, specificity, positive predictive value, and negative predictive value) and reliability (Cohen’s kappa coefficient) were evaluated for normal hearing, hearing loss, hearing impairment, permanent standard threshold shift (PSTS), temporary standard threshold shift (TSTS), and noise-induced hearing loss (NIHL).

The sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) for normal hearing, hearing loss, PSTS, and TSTS were all 100.0%, with perfect agreement (κ = 1.000). For hearing impairment, sensitivity, specificity, PPV, and NPV were 100.0%, 99.5%, 99.5%, and 100.0%, respectively, with almost perfect agreement (κ = 0.994). For NIHL, sensitivity, specificity, PPV, and NPV were 90.3%, 98.0%, 95.0%, and 96.1%, respectively, also showing almost perfect agreement (κ = 0.922).

AUDEXCEL demonstrated high validity and reliability in replicating expert diagnoses and may serve as a supportive diagnostic aid for ONRHD. However, its use should be complemented by clinical and occupational exposure assessments due to the inherent complexity of diagnosing ONRHD.